Social Studies

The Helpfulness of Crowds

 , , , ,

For the past few weeks I’ve been trying out a few of the collaborative recommendation sites that are  currently on the scene. My interest was initially piqued by an NYC startup or two, then I learned about Aardvark, now part of Google and one of the most successful of the purely crowdsourced Q&A sites.

You ask a question  and one of Aardvark subscribers is likely to have an answer. I was quite impressed that I got a quick and meaningful response about a good Spanish white wine to match with seafood.

Not all of the cloud-based oracles work this way.  Many instead rely on machine-learning techniques (decision trees, Bayesian classifiers, clustering, …) in which the mileage of your answers will vary based on the terrain of the data.

Then I learned about a mega success story from MetaFilter, a community weblog, in which two young Russian women were saved from entering the sex trade by MetaFilter members.

There I was on Friday trying to learn more about the new Google Prediction APIs when the MetaFilter story about the Russian woman started to spread through the Internet.

A manually drawn decision tree diagram drawn o...
Image via Wikipedia

With help from the crowd, the women were successfully steered away from working at a Russian “club” in Brooklyn.  You can read Newsweek’s summary to learn about one of the many heroes that were involved.

One lesson for me was that out-of-style community sites perform far better than you would expect. I wouldn’t dream of posting this type of question on Aardvark or Let Simon Decide or leave the answer up to machine learning algorithms that were designed for recommending books or movies.

Community sites still matter. And there’s nothing more powerful than getting your question or problem in front of the eyes of a human expert.

One of the interesting characteristics of communities like MetaFilter is their small world properties (or “6 degrees of separation”).

In other words, a query is likely to find an appropriate responder after just a making a few hops from the originator. I believe that was case with the MetaFilter-Russian-women query as community members followed each others’ posts and responded, thereby spreading the query to friends and friends of friends, etc.

Interesting and incredibly satisfying to see the community response as it played out in real-time.

Related articles by Zemanta
Reblog this post [with Zemanta]